dhis2-users team mailing list archive

[Jason Pickering <jason.p.pickering@xxxxxxxxx>]

Hi Mark,
The other feature you can leverage here is the "Aggregation levels"
http://dhis2.org/doc/snapshot/en/user/html/ch04.html#d5e495. In Zambia
we use a catchment population at the facility level (to be able to
take into these factors which you mention) but from the district level
onwards, indicators are calculated with official district populations.
Facility level indicators are calculated with catchment populations,
but these catchment populations do not necessarily add up to the total
of the district population. Proper configuration of the "Aggregation
levels" can allow you to implement this type of situation.

Best regards,
Jason


On Sun, Mar 11, 2012 at 2:47 AM, Mark Spohr <mhspohr@xxxxxxxxx> wrote:
> Thanks for this idea.  In some ways this could be a good solution to the
> sticky problem of determining catchment areas.  It is always difficult to
> get local population figures and to then calculate the catchment area for
> each facility.  However, this solution of prorating the population by common
> OP visits would get around this problem.  It would also account for things
> such as a particular clinic being more "accessible" in spite of greater
> distance, etc. Often the boundaries of catchment areas are drawn along
> political lines which may not correspond to natural patterns of access (such
> as a facility being near a market).
> On the other hand, it will mask facilities which people avoid because they
> are doing a poor job or don't have proper resources.
>
> Fortunately, we have a recent census.  However, we don't have low level
> village numbers yet so it will be interesting to try out this prorated
> scheme.
>
> Regards,
> Mark
>
>
> On Fri, Mar 9, 2012 at 12:27 AM, Johan Sæbø <johansa@xxxxxxxxxx> wrote:
>>
>> Just to add to this:
>>
>> If you don't have pop data below a certain level, but plenty of other data
>> in DHIS, you can always estimate population by looking at service data. This
>> needs some work, and is dependent on decent (and uniform) data completeness.
>>
>> Example: Region 1, with a known population, and under it: District A, B, C
>> with unknown population.
>>
>> Find a basket of data elements that should reflect normal use of the
>> primary health services (and which you have plenty data for, complete and
>> over time), typically including "headcount", ANC visits, Immunization, etc.
>> Find the totals for A, B, C, and their respective share of Region 1's total.
>> Then you can split Region 1 population into approximate district shares.
>>
>> This does not represent the actual target population, but those that use
>> the facilities included. At facility level, this can have some importance as
>> it hides the information about preferance of facilities, i.e. we fail to
>> find those facilities with low utilization compared to their real catchment
>> population.
>>
>> Johan
>>
>>
>>
>>
>> On 09.03.2012 06:31, Jason Pickering wrote:
>>>
>>> Yes, that is the point. Each orgunit should have a population for a
>>> specified time period (for instance, yearly population values).  So,
>>> the indicator formula would look something like this
>>>
>>> Numerator = Whatever data element
>>> Denominator = Population
>>>
>>>
>>> If you enter at level 3, you will not be able to get level 4
>>> indicators, but values will be able to be aggregated up the hierarchy.
>>>
>>>
>>>
>>> On Fri, Mar 9, 2012 at 7:20 AM, Mark Spohr<mhspohr@xxxxxxxxx>  wrote:
>>>>
>>>> Except each org unit will have population.
>>>> How do you specify which to use in an indicator? ... will these be
>>>> aggregated? ... what if no low level numbers?
>>>>
>>>> Mark Spohr MD
>>>>
>>>> On Mar 8, 2012 9:10 PM, "Jason Pickering"<jason.p.pickering@xxxxxxxxx>
>>>> wrote:
>>>>>
>>>>>
>>>>> Create a new data element called "Population". Probably want to use
>>>>> "Average" as an aggregation operator here. Add it to a dataset like
>>>>> any normal data element and enter the data through the data entry
>>>>> screen. :)
>>>>>
>>>>>
>>>>> On Fri, Mar 9, 2012 at 7:01 AM, Mark Spohr<mhspohr@xxxxxxxxx>  wrote:
>>>>>>
>>>>>> I am probably missing something obvious here but where do you enter
>>>>>> population numbers for organization units to use as denominators for
>>>>>> indicators?
>>>>>>
>>>>>> Mark Spohr MD
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Mailing list: https://launchpad.net/~dhis2-users
>>>>>> Post to     : dhis2-users@xxxxxxxxxxxxxxxxxxx
>>>>>> Unsubscribe : https://launchpad.net/~dhis2-users
>>>>>> More help   : https://help.launchpad.net/ListHelp
>>>>>>
>>>
>>> _______________________________________________
>>> Mailing list: https://launchpad.net/~dhis2-users
>>> Post to     : dhis2-users@xxxxxxxxxxxxxxxxxxx
>>> Unsubscribe : https://launchpad.net/~dhis2-users
>>> More help   : https://help.launchpad.net/ListHelp
>>
>>
>>
>> _______________________________________________
>> Mailing list: https://launchpad.net/~dhis2-users
>> Post to     : dhis2-users@xxxxxxxxxxxxxxxxxxx
>> Unsubscribe : https://launchpad.net/~dhis2-users
>> More help   : https://help.launchpad.net/ListHelp
>
>
>
>
> --
> Mark Spohr, MD
>
>
> _______________________________________________
> Mailing list: https://launchpad.net/~dhis2-users
> Post to     : dhis2-users@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~dhis2-users
> More help   : https://help.launchpad.net/ListHelp
>